Cerebras Inference: trial credits, limits and failover

What is free on Cerebras

Cerebras does not currently offer a free tier. New accounts receive $5 in credits after adding a verified payment method, and the credits expire 30 days later. When they run out or expire, API and Playground access stop until you buy credits.

Cerebras free-tier limits

On the trial, gpt-oss-120b and qwen-3.8-27b each allow 5 requests per minute, 90,000 tokens per minute (30,000 uncached), and 1 million tokens per hour and per day. Cerebras calls these a general overview; your account's Limits section is authoritative.

Get a Cerebras API key

Sign up, add a verified payment method, then open API Keys. Without a payment method, Playground and API access stay inactive.

https://cloud.cerebras.ai

OpenAI compatibility

Cerebras is compatible with OpenAI client libraries for many common workflows. Documented gaps: n must be 1, no tool streaming, images only as base64 data URIs, and tools cannot be combined with response_format on gpt-oss-120b.

https://api.cerebras.ai/v1

Data and privacy

Cerebras' privacy policy says it does not retain inputs and outputs of its inference services and deletes their logs when no longer needed.

Through Ficelle, your Cerebras key stays on your machine and prompts go straight to Cerebras: there is no Ficelle proxy in between.

Before you rely on it

• Rate-limit mechanics rechecked on 6 October 2026: uncached and total token buckets are enforced independently. Input tokens plus max_completion_tokens reserve quota before processing; choose an output limit that fits the task.
• Trial credits expire after 30 days and need a verified payment method.
• Trial context is smaller than paid: about 64,000 to 65,000 tokens instead of 128,000 to 131,000.
• The Ficelle adapter uses free-quota eligibility metadata. Its presence in the provider pack does not turn current Cerebras trial credits into renewable free access; check your account and billing settings before enabling it.

Use Cerebras with automatic failover

Ficelle can include Cerebras behind your local OpenAI-compatible endpoint when the integration is enabled and its models pass free-access and capability checks. Cerebras is listed in Ficelle Pro's maintained provider pack. Check your installed configuration, credentials and account entitlement before relying on it. Another eligible candidate must be available for failover to succeed. Local routing guards do not control upstream billing when an account permits paid usage beyond its free allocation. See strict-zero routing and account limits.

# Cerebras ships in Ficelle Pro's maintained provider pack.
ficelle doctor --text          # check this install before configuring provider access

from openai import OpenAI

client = OpenAI(base_url="http://127.0.0.1:8646/v1", api_key="ficelle-local")

# Ficelle selects among configured, enabled, eligible models.
# If none can serve the request, it returns an error.
client.chat.completions.create(
    model="ficelle/auto-tools",
    messages=[{"role": "user", "content": "..."}],
)

See what Ficelle Pro includes →

Pick a model type

Choose a model type for your request. Eligibility depends on the installed provider configuration, catalog data, free-access evidence and capability checks. A catalog listing alone does not confirm that your account can call a model.

• Tool calling
• JSON & structured output
• Fast, low-latency
• Reasoning
• Coding assistants
• Agent orchestration
• Long context
• Context compression
• Vision
• Audio input
• Video input
• Multimodal

Questions